Andy Mcloughlin

We believe in the trinity: model, inference stack, hardware. Companies that focus on a single component of this trinity lack sovereignty and are constrained by the architectural choices made by others. Most labs treat on-device models as scaled-down versions of their cloud-focused cousins. But LLM architectures that evolved for the cloud are not well-suited to on-device setups. Cloud LLMs operate in the arithmetic-bound regime. Mainstream architectures aim to maximise total token throughput by reducing the amount of computation performed per request, and they treat device memory as an unlimited resource. But for on-device deployment, memory is the main bottleneck, both in terms of throughput and the size of the resident set. When designing our on-device architecture, we focus on three core objectives: increasing the arithmetic intensity of the decoding stage, reducing the size of the resident set, and maximally utilising the GPU neural accelerators. This leads us to models that differ from traditional autoregressive transformers in a number of meaningful ways.

Your AI and cloud contracts deserve more than a spreadsheet. Duckbill helps you negotiate, manage, and forecast your compute across every cloud, neocloud, and inference provider with confidence.

Unkey is an open source API management platform that helps developers secure, manage, and scale their APIs. Unkey has built-in features that can make it easier than ever to provide an API to your end users, including: - Per key rate limiting - Limited usage keys - Time-based keys - Per key analytics

Oxen.ai is a developer platform for fine-tuning and evaluating models on your own data. Oxen.ai gives you the tools and infrastructure to build the best model for your use case. The platform makes it easy to spin up serverless GPU infrastructure to train models or run inference. When you are done, the datasets, models weights, and code are all versioned and stored in an Oxen.ai repository for easy iteration, comparison, and collaboration. Recognized by Meta in the Inaugural cohort of the Llama Startup Program.

3rd party scripts have unlimited reign in the browser of your users. We change that. Monitoring, Securing & Optimizing 3rd party scripts.

Acquired by F5, SurePath AI is a security and governance platform built for an AI-everywhere world. As AI agents act, developers code with copilots, and organizations embed AI into products, traditional security tools fall short. SurePath AI governs AI usage at the network layer across public and private models, AI coding assistants, homegrown applications, and autonomous agents. Through real-time interception, model routing, and policy enforcement, SurePath AI provides full visibility and control without slowing teams down. It secures AI access, restricts interactions to approved systems, detects malicious or risky behavior, and protects sensitive data before it reaches a model. SurePath AI ensures AI accelerates business outcomes, not risk.

Data transformation and process automation, seamlessly embedded in your product. Deliver an exceptional user experience while automating complex workflows. bem empowers engineering teams to transform any data into the exact schema they need, effortlessly managing complex, unstructured, and legacy formats. With bem, you can integrate AI-driven data transformation directly into your product, allowing you to streamline processes and enhance your platforms capabilities for your users. We partner with software companies in industries like supply chain, logistics, healthcare, and finance to automate the transformation of thousands of emails, documents, and data points. Using state-of-the-art, enterprise-grade AI, bem can be embedded directly into your product, enabling you to offer seamless, real-time data transformation as part of your own platform. By eliminating manual data processing bottlenecks, we help businesses operate with speed, accuracy, and efficiency. Whether its integrating legacy systems or automating complex workflows, bem delivers scalable, reliable AI solutions that power the future of business operations.

Lunar.dev is the infrastructure layer for governing outbound API traffic in the age of agentic AI. Built for engineering teams running at scale, Lunar.dev provides a production-grade gateway to manage and optimize traffic from AI-powered applications, including LLM calls, third-party APIs, and Model Context Protocol (MCP) interactions. Our platform gives your team visibility, policy enforcement, and active controls over API consumption. It helps you move from experimentation to safe, scalable deployment. Whether youre looking to implement token-aware rate limits, cost-based routing, or access controls for autonomous agents, Lunar.dev gives you the tools to get there.

Anything turns English into apps. Its the AI product engineer for the next wave of entrepreneurs. Everything you need to make money on the internet built in. Our team is small and talent dense by design. We’ve built things people love. So we care that Anything lets anyone build things people love.

DualBird is building the future of data infrastructure for AI, bringing cloud-native hardware acceleration to data processing. Our first product accelerates Apache Spark and Iceberg workloads on AWS by 10-30x while reducing cloud costs by 50%-90% and eliminating manual performance tuning and calibration. Our acceleration operates seamlessly under existing Spark deployments such as Amazon EMR or Spark on EKS, requiring no framework migration, code changes, or configuration tuning. Built by veterans of AWS, Databricks, NVIDIA, and leading semiconductor companies, and backed by Lightspeed, Bessemer and others, were bringing hardware acceleration to the largest compute domain that still runs on general-purpose CPUs.

Stark helps companies supercharge accessibility from months to minutes with a simple end-to-end workflow. We connect to the tools you use and love, offer automated intelligent analysis, and seamless fixes from design to code.

Nuon is a developer platform that lets you deploy into customer accounts in minutes, not months. Nuon gives you the tools to ship BYOC like it’s SaaS, and debug when things go wrong. Build applications that directly integrate with customer networks, sovereign data and systems.

Sperta helps financial services and tech companies automate decisions and manage fraud, credit, and compliance risks.

Common Paper is the structured contract platform that helps you close faster by building, negotiating, and signing contracts that both sides can make sense of.

Unified CPQ, Billing, and Metering Platform for the Modern B2B SaaS Enterprise. We support all sales channels (including marketplaces) and all business models (including real-time usage billing).

We believe in the trinity: model, inference stack, hardware. Companies that focus on a single component of this trinity lack sovereignty and are constrained by the architectural choices made by others. Most labs treat on-device models as scaled-down versions of their cloud-focused cousins. But LLM architectures that evolved for the cloud are not well-suited to on-device setups. Cloud LLMs operate in the arithmetic-bound regime. Mainstream architectures aim to maximise total token throughput by reducing the amount of computation performed per request, and they treat device memory as an unlimited resource. But for on-device deployment, memory is the main bottleneck, both in terms of throughput and the size of the resident set. When designing our on-device architecture, we focus on three core objectives: increasing the arithmetic intensity of the decoding stage, reducing the size of the resident set, and maximally utilising the GPU neural accelerators. This leads us to models that differ from traditional autoregressive transformers in a number of meaningful ways.

Your AI and cloud contracts deserve more than a spreadsheet. Duckbill helps you negotiate, manage, and forecast your compute across every cloud, neocloud, and inference provider with confidence.

Unkey is an open source API management platform that helps developers secure, manage, and scale their APIs. Unkey has built-in features that can make it easier than ever to provide an API to your end users, including: - Per key rate limiting - Limited usage keys - Time-based keys - Per key analytics

Oxen.ai is a developer platform for fine-tuning and evaluating models on your own data. Oxen.ai gives you the tools and infrastructure to build the best model for your use case. The platform makes it easy to spin up serverless GPU infrastructure to train models or run inference. When you are done, the datasets, models weights, and code are all versioned and stored in an Oxen.ai repository for easy iteration, comparison, and collaboration. Recognized by Meta in the Inaugural cohort of the Llama Startup Program.

3rd party scripts have unlimited reign in the browser of your users. We change that. Monitoring, Securing & Optimizing 3rd party scripts.